OmniSci maintains a GitHub library, mapd-core, for the OmniSciDB SQL engine. Developers can send SQL queries to mapd-core, by using our connector components like mapd-connector, and use the result with any data visualization tool. OmniSci utilizes the Vega backend rendering engine to generate geospatial images computed on the GPU from their large datasets. Developers simply pass the correct JSON string that conforms to the Vega specification, and OmniSci returns an image
OmniSciDB is able to accelerate a variety of data visualization, BI and GIS tools by executing queries orders of magnitude faster than legacy systems. OmniSci Render can also be used to serve large-scale geo visualizations to third-party tools, enhancing their "at-scale" geospatial capabilities.
OmniSciDB integrates seamlessly with the broader data science and machine learning ecosystem. Python developers can leverage the native Python DBAPI client, JupyterLab integration, or Ibis driver, which provides the expressivity of Pandas but at massive scale. Machine learning practitioners can tap the native Apache Arrow support in OmniSci to push query results directly from OmniSci into their algorithms of choice, such as Tensorflow or H2O’s XGBoost, all without the data ever leaving the GPU. This makes it easier and faster to do pre-processing, feature engineering, modeling, and comparison of predictions to outcomes.